replace manual code lists with generated lists

* Manually building hashes for code lists was leading to out of data
  lists
* switching to loading all the lists from generated data files
* Thanks to Vivek for prompting me to get around to this

bin/onix_extract_codelists

@@ -0,0 +1,16 @@
+#!/usr/bin/ruby
+# coding: utf-8
+
+USAGE = "./onix_extract_codelists ONIX_BookProduct_CodeLists.xsd some_dir"
+
+require 'rubygems'
+$LOAD_PATH.unshift(File.dirname(__FILE__) + '/../lib')
+require "onix"
+
+unless ARGV.size == 2
+  puts USAGE
+  exit(1)
+end
+
+extractor = ONIX::CodeListExtractor.new(ARGV.shift)
+extractor.run(ARGV.shift)

lib/onix.rb

@@ -2,6 +2,7 @@
 
 require 'bigdecimal'
 require 'cgi'
+require 'singleton'
 require 'roxml'
 require 'andand'
 
@@ -82,18 +83,11 @@ def self.two_digit
 require File.join(File.dirname(__FILE__), "onix", "reader")
 require File.join(File.dirname(__FILE__), "onix", "writer")
 
-# lists
-require File.join(File.dirname(__FILE__), "onix", "lists", "product_form")
-require File.join(File.dirname(__FILE__), "onix", "lists", "product_availability")
-require File.join(File.dirname(__FILE__), "onix", "lists", "country_code")
-require File.join(File.dirname(__FILE__), "onix", "lists", "language_code")
-require File.join(File.dirname(__FILE__), "onix", "lists", "language_role")
-require File.join(File.dirname(__FILE__), "onix", "lists", "notification_type")
-require File.join(File.dirname(__FILE__), "onix", "lists", "product_form_detail")
-
 # product wrappers
 require File.join(File.dirname(__FILE__), "onix", "simple_product")
 require File.join(File.dirname(__FILE__), "onix", "apa_product")
 
 # misc
+require File.join(File.dirname(__FILE__), "onix", "lists")
 require File.join(File.dirname(__FILE__), "onix", "normaliser")
+require File.join(File.dirname(__FILE__), "onix", "code_list_extractor")

lib/onix/code_list_extractor.rb

@@ -0,0 +1,69 @@
+# coding: utf-8
+
+module ONIX
+
+  # A utility class that processes the code list XSD from the ONIX spec and
+  # creates a set of TSV files. The generated files are used by this library
+  # to make hashes of the code lists available to users.
+  #
+  class CodeListExtractor
+
+    # Creates a new extractor. Expects the path to a copy of the code lists
+    # file from the spec (called ONIX_BookProduct_CodeLists.xsd on my system).
+    #
+    def initialize(filename)
+      raise ArgumentError, "#{filename} not found" unless File.file?(filename)
+
+      @filename = filename
+    end
+
+    # generate a set of TSV files in the given directory. Creates the directory
+    # if it doesn't exist and will overwrite existing files.
+    #
+    def run(dir)
+      FileUtils.mkdir_p(dir) unless File.directory?(dir)
+
+      each_list do |number, data|
+        #puts number
+        file = number.to_s.rjust(3, "0") + ".tsv"
+        path = File.join(dir, file)
+        File.open(path, "w") { |f| f.write data}
+      end
+    end
+
+    private
+
+    def data
+      @data ||= File.open(@filename) { |f| f.read }
+    end
+
+    def document
+      @document ||= Nokogiri::XML(data)
+      @document.remove_namespaces! if @document.namespaces.size > 0
+      @document
+    end
+
+    def each_list(&block)
+      document.xpath("//simpleType").each do |node|
+        list_name   = node.xpath("./@name").first.value
+        list_number = list_name[/List(\d+)/,1].to_i
+        if list_number > 0
+          yield list_number, list_data(list_number)
+        end
+      end
+    end
+
+    def list_data(num)
+      str   = ""
+      nodes = document.xpath("//simpleType[@name='List#{num}']/restriction/enumeration")
+      nodes.each do |node|
+        code  = node.xpath("./@value").first.value
+        desc  = node.xpath("./annotation/documentation").first.text
+        ldesc = node.xpath("./annotation/documentation").last.text
+        str += "#{code}\t#{desc}\t#{ldesc}\n"
+      end
+      str
+    end
+
+  end
+end

lib/onix/lists.rb

@@ -0,0 +1,122 @@
+# coding: utf-8
+
+module ONIX
+
+  class Lists
+    include Singleton
+
+    # retrieve a hash with the specified code list
+    #
+    #  ONIX::Lists.list(7)
+    #  => { "BB" => "Hardback", ... }
+    #
+    def self.list(number)
+      self.instance.list(number)
+    end
+
+    # Shortcut to retrieve a common code list
+    #
+    def self.audience_code
+      self.instance.list(28)
+    end
+
+    # Shortcut to retrieve a common code list
+    #
+    def self.contributor_role
+      self.instance.list(17)
+    end
+
+    # Shortcut to retrieve a common code list
+    #
+    def self.country_code
+      self.instance.list(91)
+    end
+
+    # Shortcut to retrieve a common code list
+    #
+    def self.language_code
+      self.instance.list(74)
+    end
+
+    # Shortcut to retrieve a common code list
+    #
+    def self.language_role
+      self.instance.list(22)
+    end
+
+    # Shortcut to retrieve a common code list
+    #
+    def self.notification_type
+      self.instance.list(1)
+    end
+
+    # Shortcut to retrieve a common code list
+    #
+    def self.product_availability
+      self.instance.list(65)
+    end
+
+    # Shortcut to retrieve a common code list
+    #
+    def self.product_form
+      self.instance.list(7)
+    end
+
+    # Shortcut to retrieve a common code list
+    #
+    def self.product_form_detail
+      self.instance.list(78)
+    end
+
+    # return a hash with the data for a single code list.
+    #
+    # number should be a fixnum specifying the list to retrieve
+    #
+    #  ONIX::Lists.instance.list(7)
+    #  => { "BB" => "Hardback", ... }
+    #
+    def list(number)
+      cache[number] ||= build_hash(number)
+    end
+
+    private
+
+    def build_hash(number)
+      val = {}
+      data(number).each_line do |line|
+        code, desc, ldesc = *line.split("\t")
+        code = code.to_i if code.to_s.match(/\d+/)
+        val[code] = desc
+      end
+      val
+    end
+
+    def cache
+      @cache ||= {}
+    end
+
+    def path(number)
+      code_dir = File.dirname(__FILE__) + "/../../data/codes"
+      filename = number.to_s.rjust(3, "0") + ".tsv"
+      File.join(code_dir, filename)
+    end
+
+    def data(number)
+      File.open(path(number)) { |f| f.read }
+    end
+
+    public
+
+    # These are here for backwards compatability with the onix gem <= 0.8.3
+    AUDIENCE_CODE        = ONIX::Lists.audience_code
+    CONTRIBUTOR_ROLE     = ONIX::Lists.contributor_role
+    COUNTRY_CODE         = ONIX::Lists.country_code
+    LANGUAGE_CODE        = ONIX::Lists.language_code
+    LANGUAGE_ROLE        = ONIX::Lists.language_role
+    NOTIFICATION_TYPE    = ONIX::Lists.notification_type
+    PRODUCT_AVAILABILITY = ONIX::Lists.product_availability
+    PRODUCT_FORM         = ONIX::Lists.product_form
+    PRODUCT_FORM_DETAIL  = ONIX::Lists.product_form_detail
+
+  end
+end